ARTICLE

https://doi.org/10.1038/s41467-019-09636-6 OPEN MKL1-actin pathway restricts chromatin accessibility and prevents mature pluripotency activation

Xiao Hu1,2, Zongzhi Z. Liu1,2,3, Xinyue Chen1,2, Vincent P. Schulz4, Abhishek Kumar 5, Amaleah A. Hartman1,2, Jason Weinstein1,2, Jessica F. Johnston 1, Elisa C. Rodriguez1, Anna E. Eastman1,2, Jijun Cheng2,6, Liz Min1, Mei Zhong1,2, Christopher Carroll1, Patrick G. Gallagher3,4,6, Jun Lu2,6, Martin Schwartz 5, Megan C. King1, Diane S. Krause1,2,7 & Shangqin Guo1,2 1234567890():,;

Actin cytoskeleton is well-known for providing structural/mechanical support, but whether and how it regulates chromatin and cell fate reprogramming is far less clear. Here, we report that MKL1, the key transcriptional co-activator of many actin cytoskeletal genes, regulates genomic accessibility and cell fate reprogramming. The MKL1-actin pathway weakens during somatic cell reprogramming by pluripotency transcription factors. Cells that reprogram efficiently display low endogenous MKL1 and inhibition of actin polymerization promotes mature pluripotency activation. Sustained MKL1 expression at a level seen in typical fibro- blasts yields excessive actin cytoskeleton, decreases nuclear volume and reduces global chromatin accessibility, stalling cells on their trajectory toward mature pluripotency. In addition, the MKL1-actin imposed block of pluripotency can be bypassed, at least partially, when the Sun2-containing linker of the nucleoskeleton and cytoskeleton (LINC) complex is inhibited. Thus, we unveil a previously unappreciated aspect of control on chromatin and cell fate reprogramming exerted by the MKL1-actin pathway.

1 Department of Cell Biology, Yale University, New Haven, CT 06520, USA. 2 Yale Stem Cell Center, Yale University, New Haven, CT 06520, USA. 3 Department of Pathology, Yale Cancer Center, Yale University, New Haven, CT 06520, USA. 4 Department of Pediatrics, Yale University, New Haven, CT 06520, USA. 5 Department of Medicine (Cardiology), Yale University, New Haven, CT 06520, USA. 6 Department of Genetics, Yale University, New Haven, CT 06520, USA. 7 Department of Laboratory Medicine, Yale University, New Haven, CT 06520, USA. Correspondence and requests for materials should be addressed to S.G. (email: [email protected])

NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6

he nucleus orchestrates characteristic gene expression The expression of many actin cytoskeletal genes is controlled Tprograms often by modulating chromatin accessibility, by the transcriptional co-activator, MKL1 (Megakaryoblastic thereby determining cellular identity. Chromatin accessi- Leukemia 1, MRTF-A), in complex with the Serum Response bility is best known to be catalyzed by biochemical activities from Factor (SRF) via binding to the CArG consensus sequence various nuclear-localized epigenetic remodeling enzymes1,2. (Supplementary Fig. 1d)19,20. The transcriptional activity of Whether the nucleus and chromatin accessibility is controlled by MKL1 is primarily controlled by its cytoplasmic-nuclear shuttling elements external to the nucleus, such as those conducting the via binding to monomeric actins21. To determine whether the biomechanical cues, is largely unexplored. subcellular localization of MKL1 changes during reprogramming, The nucleus is physically connected with the cytoskeleton via we tracked the localization of a virally expressed MKL1-GFP the linker of the nucleoskeleton and cytoskeleton (LINC) com- fusion protein22 in mouse embryonic fibroblasts (MEFs) under- plex, a highly conserved nuclear envelope bridge consisting of going reprogramming, during which all cells expressed OKSM Sun proteins and Nesprins3–5. It is known that the cytoskeleton when doxycycline (Dox) was added23. While many cells displayed and the LINC system are responsible for physically positioning nuclear MKL1-GFP at the onset of reprogramming, as expected the nucleus inside the cell and for deforming it in response to for fibroblasts grown in the presence of serum24, its subcellular mechanical signals6–9. Mechanical strains on the nucleus medi- localization became predominantly cytoplasmic when reprogram- ated by the actomyosin system could be severe enough to cause ming progressed (Fig. 1a, b), indicating reduced MKL1 activity. nuclear envelope herniation or rupture7,10–12. Strains from The altered MKL1 localization was paralleled by significant polymerized actins have also been reported to cause transcrip- reduction in F-actins, as determined by phalloidin staining tional repression13. These evidences suggest that in addition to (Fig. 1c). At gene expression level, the resulting induced regulating the physical state of the nucleus, the cytoskeleton pluripotent stem cells (iPSCs) showed reduced expression of might also be able to modify the nucleus’ biochemical state. many actin cytoskeletal genes, as well as low levels of endogenous However, the extent and nature of this modulation, as well as the MKL1 as compared to MEFs (Fig. 1d). Of note, mouse embryonic underlying mechanism remain unclear. stem cells (mESCs) displayed similarly low expression and We explored these questions using somatic cell reprogramming polymerization of actins (Fig. 1c, d), suggesting that the MKL1- into pluripotency as a model system. Pluripotent stem cells dis- actin pathway activity is low in pluripotent cells. Overall, MKL1 play highly open/accessible chromatin14,15, which can be activity and actin cytoskeletal genes are decreased during somatic experimentally induced from somatic cells of much reduced cell reprogramming into pluripotency. genomic accessibility. During reprogramming, when the tran- scription factors Oct4/Sox2/Klf4 (OSK) are first expressed in fibroblasts, they fail to bind the authentic pluripotency sites even MKL1 elevates cytoskeletal genes and blocks pluripotency.To though they are considered to possess pioneer activity16,17. The test whether the reduction in actin and related genes merely mark promiscuous binding by these pioneer factors to the somatic the cells progressing toward pluripotency, or functionally control genome suggests that chromatin accessibility might be initially cell fate transition into pluripotency, we sustained the level of constrained by mechanisms that are particularly active in somatic actin cytoskeletal genes by expressing a constitutively active cells. Here, we report that the actin cytoskeleton, and the main mutant MKL1 (caMKL1)21, or the control vector, in repro- transcription factor complex controlling its abundance, MKL1/ grammable MEFs that also contain an Oct4:GFP knock-in SRF, limits cell fate reprogramming by regulating global chro- reporter25 (Fig. 2a). The choice of expressing MKL1, instead of matin accessibility. High MKL1 activity generates excessive specific actin genes, is because MKL1/SRF concertedly upregulate actins, polymerization of which leads to a significantly reduced many actin and related genes19,20, as is the case represented by nuclear volume via a mechanism involving the LINC complex. the slow cycling cells (Supplementary Fig. 1b). A constitutively Within the small nucleus, chromatin accessibility is impaired and nuclear localized MKL1 was necessary because the wild type endogenous pluripotency fails to establish. Overall, we propose (WT) protein is inhibited by G-actin21, resulting in cytoplasmic that the actin cytoskeleton is capable of constraining global localization (Fig. 1a, b) and ineffective upregulation of actin genes chromatin accessibility. The highly accessible pluripotent genome as reprogramming progresses. The MKL1 mRNA level yielded by is accommodated by a weak actin cytoskeleton. ectopic expression was comparable to that of bulk MEFs (Fig. 2b). In the presence of Dox, both caMKL1-expressing and empty vector control MEFs gave rise to alkaline phosphatase (AP)+ Results colonies of ESC/iPSC-like morphology (Fig. 2c), indicating suc- Reprogramming is accompanied by reduced actin-MKL1 cessful early reprogramming. However, unlike control iPSCs, activity. Our previous work revealed that somatic cells with an caMKL1-expressing cells failed to activate the endogenous plur- ultrafast cell cycle are efficiently reprogrammed via ectopic ipotency reporter Oct4:GFP, even after prolonged Dox treatment expression of Oct4/Sox2/Klf4/Myc (OSKM), a property that (Fig. 2d, Supplementary Fig. 1e). They remained strictly depen- allows for their prospective isolation18. The fast cycling cells were dent on exogenous reprogramming factors, as Dox withdrawal morphologically distinct as compared to their slower cycling induced their rapid reversion to a fibroblastic morphology and counterparts (Supplementary Fig. 1a). While the slow cycling cells loss of AP expression (Fig. 2e), suggesting that sustained MKL1 had a typical fibroblastic appearance, the fast cycling cells activity inhibits the activation of mature pluripotency. appeared light-reflective and minimally spread (Supplementary Examination of reprogramming from additional somatic cell Fig. 1a). This morphological distinction suggests underlying dif- types, i.e. hematopoietic cells, confirmed the inhibitory effect by ferences in the level and/or conformation of their cytoskeletal caMKL1 (Supplementary Fig. 2). Specifically, granulocyte and components. Indeed, the fast cycling cells displayed reduced macrophage progenitors (GMPs), which reprogram efficiently as expression in many actin and related genes (Supplementary reported before18,26,27, naturally express lower actin-MKL1 Fig. 1b), but not in tubulin genes (Supplementary Fig. 1c)18, pathway activity as compared to hematopoietic stem cells revealing a specific correlation with the actin cytoskeletal system. (defined as Lin-Kit + Sca + or LKS cells), which are not efficient Thus, we investigated the role of the actin-based cytoskeleton in in reprogramming despite being more primitive in their reprogramming. developmental stage (Supplementary Fig. 2a). caMKL1 expression

2 NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6 ARTICLE

a Day 1 Day 2 Day 3 Day 4 Day 10 colony MKL1-GFP Bright field

MKL1-GFP fusion during reprogramming of reprogrammable MEFs

b c MEFs D2 D4 D6 = 228 = 58 = 136 = 487 = 563 n n n n n 120 Nuclear MKL1 Cytoplasmic MKL1 80

40 D8 D10 D25 iPS ES % of cells

0 1234D10 colony Reprogramming days

d MEF iPS ES 2.50 0.015

1.25 0.010

0.02 0.005 Expression

relative to Gapdh 0.00 0.000 Acta2 Actb Actc1 Actg2 Tagln MKL1

Fig. 1 Somatic cell reprogramming to pluripotency is accompanied by reduced actin-MKL1 pathway activity. a Time-course imaging of MKL1-GFP localization in MEFs undergoing reprogramming. Representative images acquired daily (day 1–4) during early reprogramming and on day 10 for a colony with iPS-like morphology. Note MKL1-GFP gradually localizes to the cytoplasm as reprogramming progresses. Scale bar: 50 μm. b Quantification of the % of cells displaying nuclear- or cytoplasmic-localized MKL1-GFP in a. Images from three independent experiments were analyzed, and data were pooled from the indicated numbers of cells for each condition. c Representative confocal images of reprogramming cultures stained with phalloidin (F-actin) through the time course (MEF, day 2, day 4, day 6, day 8, day 10, and day 25). A typical ESC colony was stained as a control. Scale bar: 20 μm. d Realtime QPCR analyses of the mRNA levels of actin related genes, as well as MKL1 itself in MEFs, iPSCs and ESCs. Data are representative of three independent experiments. Error bars denote standard deviation of triplicate samples (n = 3) in each experiment

(mCherry+) (Supplementary Fig. 2b) similarly and potently Data 2), primarily to CArG motif sequences (Supplementary prevented the appearance of Oct4:GFP+ cells from reprogram- Fig. 3d). These data confirm that a main molecular consequence mable GMPs (Supplementary Fig. 2c, d, e). Thus, MKL1 activity of caMKL1 expression is the elevation of many actin cytoskeletal prevents activation of mature pluripotency from multiple somatic components. cell types. To position the cellular stage onto the molecular roadmap of To uncover the molecular consequences of MKL1 activity, we pluripotency induction30, we compared the differentially performed mRNA-seq and compared reprogrammable MEFs expressed genes between day 25 control- and caMKL1- expressing control vector, or caMKL1, at early (Day 6) or late expressing cells to gene signatures of cells undergoing repro- (Day 25) time points (Fig. 2f, g). As expected, the most gramming (Supplementary Fig. 3e, f). Gene Set Enrichment significantly up-regulated gene ontology (GO) categories in Analysis (GSEA) showed that caMKL1-expressing cells failed to caMKL1-expressing cells were actin-related biological pathways up-regulate many late pluripotency genes, such as Esrrb and (Fig. 2f, g and Supplementary Fig. 3a, Supplementary Data 1), endogenous Sox2, but sustained the expression of many others confirming MKL1 activity. To examine whether this altered gene that should be downregulated during reprogramming (Supple- expression program relates to the transcriptional activity of mentary Fig. 3e, f). Consistent with the gene expression data, MKL1, we determined the binding of its transcriptional partner ChIP-seq analyses of H3K4me3, H3K27ac and H3K27me3 SRF by chromatin immuno-precipitation followed by sequencing confirmed the lack of active histone marks at the endogenous (ChIP-seq), since a ChIP-grade antibody for MKL1 is not pluripotency loci in caMKL1-expressing cells (Supplementary commercially available while SRF genome binding profile can be Fig. 3g). While Oct4 bound to its bona fide target regions in the used to gauge MKL1 binding28,29. As expected, caMKL1 genome of control iPSCs, such as the Oct4 enhancer, it was expression resulted in significantly stronger SRF binding to many depleted from these regions in caMKL1-expressing cells (Fig. 2h, cytoskeletal genes (Supplementary Fig. 3b, c, Supplementary Supplementary Data 2). These caMKL1-expressing, AP+ cells

NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6

a b MKL1 RNA pMSCV caMKL1 mCherry 0.015 IRES NC 0.010 MKL1 0.005 RPEL domain Col1a1 TetO Oct4 Klf4Sox2 c-Myc Expression 0.000 relative to Gapdh relative caMKL1 F2A IRES E2A Δ (NLS- RPEL) Rosa26 M2-rtTA MEF

NLS Control Endogenous Oct4 IRES caMKL1 GFP Reprogramming day 25 c d e Day 10 + Dox – Dox Control caMKL1 Control caMKL1 400 40 Control caMKL1 300 30

Control 200 20 100 10 Oct4:GFP % 0 0 AP+ colonies/4000 cells caMKL1 Control D12 D16 D20 D25 D30 caMKL1 Reprogramming days

fg h Control caMKL1 Control iPS Red: padj <= 0.05, Red: padj <= 0.05 iPS D25 fold change >= 2 fold change >= 2 caMKL1 D25 10 Myh11 10 Actc1 10 caMKL1 caMKL1 2.0 Actc1 Actg2 Myh11 2 5 Actg2 5 Acta2 1.5 MKL1 Acta2 Acta1 20 Actb 1.0 0

0 Input % EpCam Sall4 0.5 Nr5a2 −5 −5 Dppa2 0.0 Utf1 10 Fold change (Log2) Fold change (Log2) Dppa4 Esrrb Control Control Dppa3 Zfp42 Dppa5 −10 −10 Oct4-occupied 1 100 10,000 1 100 10,000 regions in ctrl iPS Oct4 enh.

Mean expression Mean expression 0 Gapdh pro. Day 6 Day 25 –5k 0 +5k

Fig. 2 MKL1 drives actin cytoskeletal gene expression and blocks mature pluripotency. a Left: schematic diagram of constitutively active MKL1 (caMKL1), with the RPEL domain deleted and replaced with a nuclear localization signal (NLS). Right: experimental design illustrating a retroviral vector expressing caMKL1-IRES-mCherry to be transduced into reprogrammable MEFs, which express Col1a:OKSM, Rosa26:M2rtTA and Oct4:GFP. The co-expressed mCherry from the vector was used to isolate the transduced cells for all further experiments. b Realtime QPCR analysis of total MKL1 mRNA level in primary MEFs, as compared to the cells at reprogramming day 25 coexpressing either a control vector or caMKL1. n = 3 for MEF and control; n = 4 for caMKL1- overexpressing condition. Error bars denote standard deviation. c AP stained reprogramming cultures on day 10, coexpressing either control vector or caMKL1. The number of AP+ colonies derived from 4,000 reprogrammable MEFs are quantified. d Percentage of Oct4:GFP+ cells arising from reprogramming MEFs coexpressing control vector or caMKL1, as determined by FACS. Cells were trypsinized and analyzed at indicated time points starting from day 10. e AP stained reprogramming cultures from control- or caMKL1-expressing reprogramming MEFs on day25, when cells were passaged and replated into continued cultures with or without Dox. Bottom panel shows colonies at higher magnification. Note that no AP+ colonies could be detected in the caMKL1-expressing cultures without Dox. f, g MA plots of differentially expressed genes between control- and caMKL1-overexpressing cells at reprogramming day 6 (f) and day 25 (g), respectively. h Meta analysis of Oct4-bound regions in day 25 reprogramming cells expressing the control vector or caMKL1 (left). The Oct4-bound regions in control iPS cells show reduced Oct4 binding in caMKL1-expressing cells. Oct4 ChIP-qPCR confirmed reduced binding at the Oct4 enhancer region in caMKL1-blocked cells (right). Error bars denote standard deviation of triplicate samples (n = 3) in each one of three independent experiments (c, d, h) appear to be in a unique state of pre-pluripotency since Nanog intermediates17,31,32 but distinct from the Nanog+ F-class was expressed at a level comparable to control iPSCs (Supple- alternative pluripotency33. mentary Fig. 3h) and the corresponding histone marks at this The inhibition of mature pluripotency activation was not due locus were properly decorated (Supplementary Fig. 3g). These to compromised cell proliferation and/or survival, because data demonstrate that MKL1 activity prevents the extensive caMKL1-expressing MEFs gave rise to more AP+ colonies than activation of endogenous core pluripotency, yielding a cell state controls (Fig. 2c), and these caMKL1-expressing AP+ colonies that is more advanced than previously reported reprograming can be stably maintained in mESC culture conditions as long as

4 NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6 ARTICLE

a b c d Control shRNA MKL1 shRNA Control shRNA Control shRNA Control shRNA Myh11 shRNA Control shRNA + ROCKi MKL1 shRNA Fmn2 shRNA Myh11 shRNA 40 10 Arp2 shRNA Myh11 shRNA + ROCKi Arp3 shRNA – Dox 30 8 4 6 3 20 4 2 10 2 1 Oct4:GFP+ cells% Oct4:GFP+ cells% Oct4:GFP+ cells% + Dox 0 0 0 0 6 8 0 0 8 10 12 14 16 18 20 10 12 14 16 18 20 10 12 14 16 Blocked cells Blocked cells Rescuing cells caMKL1 blocked cells with shRNA (days) with shRNA (days) with ROCKi (Days)

e fgh Control shRNA Control shRNA Control shRNA + ROCKi Control Control Control shRNA + ROCKi Arp2 shRNA ROCKi ROCKi Fmn2 shRNA Arp2 shRNA + ROCKi Arp3 shRNA CK-666 CK-666 Fmn2 shRNA + ROCKi Arp3 shRNA + ROCKi CK-666 + ROCKi CK-666 + ROCKi 15 15 20 50 40 15 Drugs added 10 10 30 10 at day 10 20 5 5 5 10 Oct4:GFP+ cells% Oct4:GFP+ cells% Oct4:GFP+ cells% 0 0 0 0 Oct4:GFP+ cells% 0128141610 0 246810 02469121416 10 12 14 16 18 20 Rescuing cells Rescuing cells Blocked cells (days) Reprogramming days with ROCKi (Days) with ROCKi (days)

Fig. 3 Actin polymerization is required for caMKL1-mediated block of mature pluripotency activation. a AP staining of caMKL1-expressing cells transduced with control or MKL1-targeting shRNA, cultured with or without Dox. Staining was performed 15 days after viral transduction. b–f Percentage of Oct4:GFP+ cells emerging from blocked cells following treatment of control or MKL1-targeting shRNAs (b); control or Myh11-, Fmn2-, Arp2- or Arp3-targeting shRNAs (c); control or Myh11-targeting shRNAs, with or without ROCKi (d); control or Fmn2-targeting shRNAs, with or without ROCKi (e); control or Arp2- or Arp3-targeting shRNAs, with or without ROCKi (f). d–f ROCKi was added at day 20 of shRNA treatment. g Percentage of Oct4:GFP+ cells emerging from blocked cells following treatment of ROCKi or CK-666, alone or in combination. h Percentage of Oct4:GFP+ cells emerging from reprogrammable MEFs in the presence of indicated drugs. Cells were treated with drugs starting at reprogramming day 10. All GFP+ percentages were determined by FACS at indicated time (days), following a gating strategy as shown in Supplementary Fig. 1. All data are representative of three independent experiments. Error bars denote standard deviation of triplicated samples (n = 3) in each experiment

Dox was added (Fig. 2c, d, e). Importantly, depleting MKL1 in mature pluripotency, strongly suggesting that the elevated actin these AP+ cells resulted in abundant mature iPSCs that could be cytoskeleton likely mediates the MKL1 blocking effect of mature maintained without Dox, whereas no mature colonies formed in pluripotency. control shRNA-treated cultures (Fig. 3a, b). Since these caMKL1- To directly test whether MKL1-mediated blockade of mature expressing cells remain in a stable pre-pluripotent state, which pluripotency is afforded by the actin cytoskeleton, we depleted readily progressed toward mature pluripotency upon MKL1 regulators of actin polymerization, namely Arp2/3 and Formin 2 inhibition, we refer to them as caMKL1-blocked cells. (Fmn2)7,37, as well as one of the myosin heavy chains, Myh11 (Fig. 3c, Supplementary Fig. 5a). With varying efficiencies, caMKL1-mediated block requires actin polymerization. The knockdown of Myh11, Fmn2 or Arp2/3 in caMKL1-blocked above analyses revealed two molecular consequences of MKL1 cells all resulted in the emergence of Oct4:GFP+ cells and activity: elevated actin cytoskeletal gene expression and failed elevated the expression of endogenous mature pluripotency genes activation of mature pluripotency. Since high levels of actin (Fig. 3c). Although the number of Oct4:GFP+ cells emerging cytoskeletal components favor polymerization, we tested whether after some of these shRNA treatments appeared moderate, these actin polymerization leads to failure in mature pluripotency targeting shRNAs always yielded significantly more Oct4:GFP+ activation, first by modulating RhoA activity, a major driver of cells than control shRNA-treated cultures, which produced no actin polymerization and actomyosin contractility34,35.As Oct4:GFP+ cells at all. The effectiveness to restore mature expected, RhoA activity promoted actin polymerization, as con- pluripotency by interfering with these regulators of actin firmed by the presence of strong phalloidin staining in cells polymerization was lower than that of the positive control, when expressing a constitutively active mutant RhoA36 (Supplementary MKL1 itself was depleted (Fig. 3b). This is likely because caMKL1 Fig. 4a). This constitutively active mutant, but not wild-type or a upregulates many cytoskeletal genes, and inhibiting one at a time dominant negative RhoA, decreased the number of Oct4:GFP+ would likely not be as effective as inhibiting MKL1 itself. Indeed, colonies when expressed in reprogramming MEFs (Supplemen- the rescue effects of these shRNAs were greatly potentiated by an tary Fig. 4b). Thus, activating the actin cytoskeleton independent inhibitor (Y27632, ROCKi) of the Rho-associated protein kinase, of caMKL1 expression could also compromise the activation of which is known to activate actin polymerization (Fig. 3d, e, f and

NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6

abeControl shRNA Sun1 shRNA *** Control shRNA 100 * Control shRNA + ROCKi **** Control shRNA Sun1 shRNA 80 Sun1 shRNA Sun1 shRNA + ROCKi Sun2 shRNA 60 Sun2 shRNA Sun2 shRNA + ROCKi 2.0 30 40

1.5 Oct4:GFP+ colony 20 20 Sun2 Sun1& number / 8000 cells 1.0 shRNA Sun2 shRNA 0 10 0.5 Oct4:GFP+ cells % Oct4:GFP+ cells % 0.0 0 Sun1 Sun2shRNA shRNA 0 101214161820 0246810 Control shRNA Blocked cells Rescuing cells Sun1&Sun2 shRNA with shRNA (days) with ROCKi (days)

c WT Sun2–/– d

2.0 0.04 0.08 0.6 8 4 1500 4 100 8 80 1.5 0.03 0.06 6 3 3 6 0.4 1000 60 1.0 0.02 0.04 4 2 2 4 40 0.2 500

Expression 0.5 0.01 0.02 2 1 1 20 2 caMKL1 vs control 0.0 0.00 0.00 0.0 0 0 0 0 0 0 Endo- Esrrb Nanog Sall4 Acta2 Actb Actc1 Actg2 Tagln MKL1 Oct4

Fig. 4 LINC complex contributes to the block of mature pluripotency activation. a Percentage of Oct4:GFP+ cells emerging from blocked cells following treatment of control, Sun1- or Sun2-targeting shRNAs. b The same experiments as in a were performed with or without ROCKi, which was added on day 20 of shRNA treatment. GFP+ percentages were determined by FACS at indicated time (days). c, d Realtime QPCR analyses of selected genes in control- or caMKL1-expressing cells, derived from reprogramming WT or Sun2−/− cells on day 40. Each bar represents the ratio of the indicated gene expression levels in caMKL1- vs control-expressing cells. Endogenous pluripotency genes are shown in c; MKL1 and its target genes are shown in d. Data are representative of three independent experiments. e AP stained cultures derived from reprogramming MEFs, while treated with control shRNA, Sun1- or Sun2-targeting shRNA, or both (left). The number of Oct4:GFP+ colonies were quantified on the right. *P < 0.05; ***P < 0.001; ****P < 0.0001. All data are representative of three independent experiments. Error bars denote standard deviation of triplicated samples (n = 3) in each experiment. Statistics were performed using two-tailed unpaired t-test

Supplementary Fig. 5b, c, d). Since Arp2/3 knockdown was most in the blockade toward mature pluripotency, we inhibited com- effective at releasing the blocked cells into mature pluripotency ponents of the LINC complex. shRNAs against Sun1 or Sun2 (Fig. 3c), we tested whether a small molecule inhibitor of Arp2/3 (Supplementary Fig. 5a) each partially released the caMKL1- (CK-666) functions similarly. Indeed, CK-666 released the blocked cells into mature pluripotency (Fig. 4a), which was fur- caMKL1-blocked cells into mature pluripotency, especially in ther enhanced by co-treatment with ROCKi (Fig. 4b). Strikingly, the presence of ROCKi (Fig. 3g and Supplementary Fig. 5e). when Sun2 shRNA and ROCKi were added at the same time, Together, these data demonstrate actin polymerization mediates the rescue efficiency reached a similar level as the positive control, MKL1-imposed restriction for mature pluripotency activation. MKL1 shRNA (compare Figs. 3b and 4b). These data suggest that The potent blocking effect mediated by the actin cytoskeleton the Sun2-containing LINC complexes together with polymerized upon caMKL1 expression, and the release from the blockade actins are responsible for most of the blocking effects by caMKL1. upon their inhibition, prompted us to ask whether this actin We further tested whether caMKL1-mediated block of mature polymerization-based mechanism antagonizes mature pluripo- pluripotency activation could be bypassed if the Sun2-containing tency induction during normal reprogramming in the absence of LINC complex is genetically inactivated. As shown above, whereas caMKL1. Indeed, CK-666 and ROCKi promoted the activation of caMKL1 expression completely prevented the activation of endogenous pluripotency genes in normal reprogramming endogenous pluripotency genes, such as Oct4 and Esrrb, in Sun2 cultures when added at day 10 (Fig. 3h). These results support wild type cells, these late pluripotency genes were activated in that the abundance and polymerization state of actins regulate Sun2 KO cells9 even in the presence of MKL1 activity (Fig. 4c, d). nuclear reprogramming, and this regulatory effect is not limited Together, these data demonstrate that the LINC complex is to the blocked cell state created by caMKL1 expression. required for MKL1 to restrict mature pluripotency activation. To examine whether the LINC complex participates in the blockade of mature pluripotency in the absence of caMKL1, we LINC complex contributes to the block of mature pluripotency. inhibited either Sun1 or Sun2 during normal reprogramming. In addition to structurally supporting the cell, the actin cytos- These treatment significantly increased the number of Oct4:GFP+ keleton connects with the nucleus via a nuclear membrane- colonies (Fig. 4e). These data demonstrate that the LINC traversing molecular bridge, the LINC complex, which further complex components are involved in restricting mature plur- connects to components of the nuclear interior via lamina- ipotency activation. Importantly, these effects are not limited to associated proteins4,5. To test whether this connection is involved the blocked cell state instituted by the expression of caMKL1.

6 NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6 ARTICLE

ab c DAPI/Phalloidin Enlarged DAPI/Phalloidin Enlarged **** 1.0 **** **** 0.8

0.6 WT ES

Control iPS 0.4

Nuclear circularity 0.2 = 52 = 63 = 60 = 77 = 77 0.0 n n n n n

Δ iPS Δ/ WT ES blocked caMKL1 ControlSRF iPS caMKL1 ES caMKL1 ES

caMKL1 blocked

deNesprin 2G reporter fES Blocked cells gcaMKL1 n.s. **** ** blocked cells Actin binding Sun binding 2.0 2.0 TFP Venus Tension 1.5 1.5 sensor (TS)

Low FRET HL 1.0 1.0 blocked caMKL1 TS TFP Venus

FRET index 0.5 0.5 relative to HL = 17 = 12 High FRET = 19 = 19 = 20 n n n n 0.0 n 0.0 Headless (HL) TFP Venus HL TS HL TS

Relaxed TS iPS Control h caMKL1 blocked Control iPS SRFΔ/Δ iPS TS + CK-666 Δ / Δ iPS SRF TS+CK-666

00.4 ijWT ES caMKL1 ES Control shRNA Arp2 shRNA caMKL1 blocked cells

Fig. 5 MKL1-actin pathway regulates nuclear volume. a, b DAPI and phalloidin staining in control iPS cells and caMKL1-blocked cells (a); WT and caMKL1- overexpressing ES cells (b). Rectangular boxes indicate regions shown at higher magnification on right. Scale bar: 20 μm. c Nuclear circularity of indicated cell types. The numbers of analyzed nuclei are as indicated. ****P < 0.0001. d Electron microscopy images showing nuclear morphology in caMKL1-blocked cells, control iPS cells and SRFΔ/Δ iPS cells. Scale bar: 2 μm. e Schematics of the Nesprin-2G tension sensor (TS) and Headless control (HL). f FRET signals yielded by these constructs at the nuclear envelope were analyzed in blocked cells, or in mESCs as a control. CK-666 reduced the tension sensor FRET signals in the blocked cells. The numbers of analyzed nuclei are as indicated. n.s.: non-significant; **P < 0.01; ****P < 0.0001. g Representative heatmap images of FRET signals in caMKL1-blocked cells shown in f. h–j Representative 3D nuclear images reconstructed from DAPI stained confocal images of caMKL1-blocked cells, control iPS cells and SRF Δ/Δ iPS cells (h), WT and caMKL1-overexpressing ES cells (i), caMKL1-blocked cells treated with control or Arp2 shRNA (j). Scale bar: 10 μm. All data shown are representative of three independent experiments. Error bars denote standard deviation. Statistics were performed using two-tailed unpaired t-test

The involvement of the LINC complex components also suggests phalloidin signals (Fig. 5a and Supplementary Fig. 6a, b). Con- a potential mechanism how a hyperactive actin cytoskeleton sistently, mESCs with ectopic caMKL1 also exhibited more could be responsible for the failure of the pluripotent cell fate. abundant cytoplasmic F-actins surrounding the nuclei, while maintaining expression of pluripotency genes (Fig. 5b and Sup- plementary Fig. 6c, d, e). Of note, mESCs with high caMKL1 MKL1-actin pathway regulates nuclear state. In searching for expression could not be established, indicating high caMKL1 potential mechanism(s) how polymerized actins, together with activity might inhibit pluripotency. the LINC complex, block mature pluripotency activation, we The functional consequence of abundant F-actins toward the noticed that the nuclei of caMKL1-blocked cells were surrounded nucleus could be seen by analyzing isolated nuclei. Due to the by abundant cytoplasmic F-actins, as indicated by the strong nucleus’ viscoelastic nature38,39, isolated nuclei could temporarily

NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6 retain their shape/size following their isolation but eventually elastic spring linker45 (Fig. 5e and Supplementary Fig. 8a). The swell and lose integrity. The ability to retain their integrity is sensor is designed to have two binding domains from Nesprin- reflected as the size of the isolated nuclei at defined time following 2G: one that binds actins on the outer nuclear envelope facing the isolation, with smaller nuclei indicating better retained integrity. cytoplasmic side, and the other that binds Sun proteins in the Isolated nuclei from caMKL1-blocked cells always appeared perinuclear space between the outer and inner nuclear mem- smaller than those of control iPSCs (Supplementary Fig. 6b), and branes. The distance between the fluorophores is determined by caMKL1-expressing mESCs similarly always yielded smaller strains on the two Nesprin domains, thus leading to altered FRET nuclei (Supplementary Fig. 6d). Importantly, this caMKL1- signals. A Headless (HL) construct lacking the actin-binding induced difference was alleviated by actin de-polymerizing agent domain, which therefore exists in a relaxed configuration, serves cytochalasin D (CD)40 in caMKL1-blocked cells as well as as baseline FRET control. We found that caMKL1-blocked cells caMKL1-expressing mESCs (Supplementary Fig. 6b, d), indicat- expressing the Tension Sensor (TS) displayed significantly higher ing the involvement of actin cytoskeleton in generating the FRET signals than those expressing the HL control (Fig. 5f, g), different nuclear states. In contrast, CD treatment had no indicating that the FRET sensor was under compression in the discernable effects on control iPSC or WT mESC nuclei, caMKL1-blocked cells. Furthermore, the high FRET signal consistent with their constitutively weak actin cytoskeleton in caMKL1-blocked cells was reduced by treatment with the (Fig. 1c, d). This assessment suggests that the abundant Arp2/3 inhibitor CK-666, indicating that the compression on the polymerized actins consequent to MKL1 activity cause qualitative FRET sensor was at least partly mediated by polymerized actins differences in nuclear state. (Fig. 5f,g). In contrast, no significant difference in FRET levels We next analyzed the nucleus within intact cells. Indeed, the was seen between the TS and HL in mESCs (Fig. 5f), as expected nuclei of the caMKL1-blocked cells appeared rounder than from their constitutively weak actin cytoskeleton (Fig. 1c, d). control iPSCs, as reflected by their higher nuclear circularity These data support that the abundant actins present in caMKL1- (Fig. 5c). This difference is not consequent to the different cellular blocked cells are functionally engaged with the nucleus via the state occupied by the blocked cells, because ectopic caMKL1 LINC complex. The elevated Nesprin FRET signals in caMKL1- expression in mESCs also increased nuclear circularity (Fig. 5c). blocked cells suggest that the LINC complex is under compres- To further examine the relationship between MKL1/SRF activity sion in these cells. and nuclear circularity, we generated SRFΔ/Δ iPSCs from SRFf/f It has been observed previously that when compressed, the cells41. Consistent with the phenotype of SRF-null mESCs42, nucleus reduces its size along the direction of compression46. The SRFΔ/Δ iPSCs maintained a pluripotent gene expression profile compressed Nesprin-2G FRET sensor prompted us to examine similar to SRFf/f iPSCs (Supplementary Fig. 6f). SRFΔ/Δ iPSCs had the relationship between MKL1-actin pathway activity and a minimal F-actins with barely detectible phalloidin signals potentially compressed nucleus, which may display decreased (Supplementary Fig. 6g), as expected from a disabled MKL1/ nuclear dimensions. To do this, we acquired and reconstructed SRF transcription program, while their nuclei were shaped much 3D whole nuclei images from confocal image stacks of DAPI- more irregularly (Fig. 5c). The difference in nuclear circularity stained nuclei. As shown in Fig. 5h and Supplementary Fig. 8b, was even more pronounced when the nuclei were examined with caMKL1-blocked cells had the smallest nuclear volume (207 ± 58 higher imaging resolution by electron microscopy (Fig. 5d). μm3), while SRFΔ/Δ iPSCs had the largest (553 ± 167 μm3), with Overall, nuclear circularity increased with MKL1 activity, with the control iPSCs in between (323 ± 176 μm3), indicating an inverse highest circularity detected in caMKL1-blocked cells and the correlation between nuclear volume and MKL1 activity. In lowest nuclear circularity in SRFΔ/Δ iPSCs. These data further addition, caMKL1 expression in mESCs reduced their nuclear indicate that the nuclear state responds to MKL1/SRF activity. volume (Fig. 5i, from 352 ± 88 to 252 ± 44 μm3), suggesting Nuclear lamina is known to modulate nuclear morphology and MKL1 activity could directly reduce nuclear volume without chromatin states43,44. To determine whether nuclear lamina is causing significant changes in key pluripotency gene expression, altered in response to MKL1 activity, which in turn may mediate likely involving nuclear compression. Further, treating caMKL1- the blockade of mature pluripotency, we immunostained lamin blocked cells with Arp2-targeting shRNA restored nuclear A/C in the caMKL1-blocked cells. This revealed that lamin A/C volume (Fig. 5h, j, from 192 ± 53 to 701 ± 137 μm3). These protein was elevated in caMKL1-blocked cells (Supplementary data are in agreement with the functional rescue of caMKL1- Fig. 7a), and became detectible in mESCs when caMKL1 was blocked cells by Arp2 inhibition (Fig. 3c, f, g and Supplementary overexpressed (Supplementary Fig. 7b). However, when we Fig. 5c, d, e). Mirroring the effects of ROCKi and CK-666 overexpressed lamin A or C in reprogramming MEFs directly, in inhibiting actin polymerization and promoting mature neither had any effect on reprogramming efficiency; similar pluripotency, ROCKi and CK-666 also helped to increase nuclear numbers of Oct4:GFP+ colonies formed despite the over- volumes in late reprogramming (Supplementary Fig. 8c). Overall, expression of lamin A or C (Supplementary Fig. 7c–e). These these data are consistent with a model where nuclear volume is data suggest that although caMKL1-expressing cells have elevated regulated by Arp2-dependent actin polymerization. lamin A/C, the increased lamin A/C is likely one consequence of MKL1 activity, and has little direct antagonism to mature pluripotency activation. Nonetheless, these results further the MKL1 activity results in reduced chromatin accessibility.A observation that nuclear state responds to MKL1 activity. The direct implication of a reduced nuclear volume as seen with abundant F-actins consequent to high MKL1 activity could caMKL1-blocked cells (Fig. 5h, Supplementary Fig. 8b) is a more profoundly change nuclear state, including altered nuclear lamina compact chromatin organization, since the same amount of protein expression and nuclear morphology. genome is confined to a smaller space in 3D. Such a model could effectively explain the failure of caMKL1-blocked cells in acti- vating mature pluripotency, likely due to decreased chromatin MKL1-actin pathway regulates nuclear volume.Todefine the accessibility. To test this possibility, we first performed DNase I functional engagement between cytoplasmic actins and the digestion assay, which revealed overall reduced chromatin nucleus, most likely via the LINC complex, we took advantage of accessibility in caMKL1-blocked cells relative to control iPSCs a Nesprin-2G FRET sensor that consists of a pair of FRET- (Fig. 6a). This reduction in chromatin accessibility is echoed and generating fluorophores, TFP and venus, connected by a linear- corroborated by reduced Oct4 chromatin binding as revealed by

8 NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6 ARTICLE

abcd Control iPS Control iPS (n = 25) WT ES SRFf/f iPS caMKL1 blocked caMKL1 blocked caMKL1 ES SRFΔ/Δ iPS 0.03 0.02 1.0 (n = 25) 0.03 ****

0.02 0.02 0.5 0.01 0.01 0.01 (% of total) (% of total) (% of total) Size distribution Size distribution Size distribution Fluorescence % 0.00 0.0 0.00 0.00 2 0.6 0.1 20 40 60 80 2 0.6 0.1 2 0.6 0.1 Fragment size (kb) H1-mCherry (seconds) Fragment size (kb) Fragment size (kb)

e +: Sites with increased accessibility in caMKL1 g FDR < 0.1 –: Sites with increased accessibility in WT ES caMKL1 blocked Blocked + UNC0642 WT ES 12 4 0.03 10 2 8 0.02

0 6 0.01 (% of total) WT vs caMKL1 in accessible sites Log2 fold change 4 Size distribution

–2 Log2 normalized reads caMKL1 ES Total + +–– 0.00 024681012 WT ES 2 0.6 0.1 Log2 normalized reads caMKL1 ES Fragment size (kb)

f +: Sites with increased accessibility in SRFΔ/Δ h FDR < 0.1 –: Sites with increased accessibility in SRFf/f f/f caMKL1 blocked 6 SRF iPS 12 30 Blocked + UNC0642 Δ

/ 10 Δ 4 8 2 6 20 vs SRF

f/f 0 4 2 10 SRF Log2 fold change

–2 in accessible sites 0 Oct4:GFP+ cells% Δ/Δ Log2 normalized reads –4 SRF iPS Total + + – – 0 0 2 4 6 8 10 12 SRFf/f iPS 0 46810121416 Log2 normalized reads SRFΔ/Δ iPS Inhibitor treatment (days)

Fig. 6 MKL1 activity results in reduced chromatin accessibility. a Quantification of the size distribution of DNase I digested DNA fragments from control iPS cells and caMKL1-blocked cells. n = 3 for each genotype. b FRAP analysis of the dynamic movement of histone H10-mCherry in control iPS cell nuclei or caMKL1-blocked cell nuclei. ****P < 0.0001. n denotes the number of analyzed nuclei. Statistics were performed using two-tailed paired t-test. Error bars denote standard deviation. c, d Quantification of the size distribution of DNase I digested DNA fragments from WT and caMKL1-overexpressing ES cells (c) and SRFf/f iPS cells and SRFΔ/Δ iPS cells (d). n = 3 for each genotype. e, f MA plots of differential ATAC-seq peaks in WT vs caMKL1-expressing ES cells (e) and SRFf/f vs SRFΔ/Δ iPS cells (f). Pink dots indicate regions of significant differences in chromatin accessibility (left). The width of boxplots (right) is proportional to the number of regions of differential chromatin accessibility, showing overall lower accessibility in caMKL1-expressing ESCs (e), and overall increased accessibility in SRFΔ/Δ iPS cells (f). The center line of boxplots indicate the median value; the edges of the boxplots are at the 25th and 75th percentile; the whiskers extend to 1.5 times the interquartile range past the box. Points that are outside whiskers are shown individually. g Quantification of the size distribution of DNase I digested DNA fragments from caMKL1-blocked cells treated with or without UNC0642. n = 3 for each genotype. h Percentage of Oct4:GFP+ cells emerging from blocked cells following treatment with UNC0642. GFP+ cells were determined by FACS. All data shown are representative of three independent experiments

ChIP-seq, shown above in Fig. 2h. In addition, fluorescence overall chromatin accessibility is modulated by the MKL1/SRF recovery after photobleaching (FRAP) analysis of mCherry- activity level as assessed by DNase I accessibility. tagged histone H10 confirmed reduced chromatin dynamics in To examine the effects of MKL1/SRF activity on chromatin the blocked cells (Fig. 6b). The difference in chromatin openness accessibility in more detail, we performed ATAC-seq47 compar- is not consequent to differences in cellular states, as mESCs dis- ing the genome-wide accessibility when MKL1/SRF activity is played a reduction in chromatin openness when caMKL1 was altered. To avoid measuring the difference in chromatin expressed (Fig. 6c). Furthermore, DNase I digestion assay accessibility consequent to different cell state, we performed revealed that SRF null iPSCs displayed increased chromatin these comparisons in mESCs with or without caMKL1 expres- accessibility beyond that of the control SRFf/f iPSCs (Fig. 6d). sion, and in iPSCs with or without a functional SRF gene. Besides modulating MKL1/SRF, expression of the constitutively Consistent with the DNase I digestion assay results (Fig. 6c, d), active RhoA mutant in mature iPSCs also decreased chromatin ATAC-seq analyses confirmed globally reduced chromatin accessibility, while expression of the dominant negative RhoA accessibility in mESCs following caMKL1 expression (Fig. 6e mutant promoted it (Supplementary Fig. 9a). Taken together, the and Supplementary Fig. 9b, Supplementary Data 3), including

NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6 many Oct4 binding sites/regions (Supplementary Fig. 9c). Con- morphology, with human cells being flat monolayers of higher versely, loss of SRF in iPSCs globally increased chromatin focal adhesion signaling54. It will be important to determine accessibility, beyond the level seen with wild type iPSCs (Fig. 6f). whether modulating actin cytoskeleton-LINC complex activity Together, these data demonstrate that MKL1 activity can result in can lead to transitions between different pluripotent states in a globally less accessible chromatin conformation. human ESC/iPSCs55,56, or in deriving bona fide pluripotent stem To assess whether the reduced chromatin accessibility in cells of certain challenging model species57. Our model might also caMKL1-blocked cells was responsible for their failed activation provide new insights to biology beyond the regulation of plur- of mature pluripotency, we promoted chromatin accessibility by ipotency, such as in the emergence of aberrant cell states means independent of the MKL1-actin pathway. Specifically, we underlying the various fibrotic conditions58,59. treated the caMKL1-blocked cells with UNC0642, an H3K9me2/3 48 methyltransferase inhibitor . As expected, UNC0642 treatment Methods increased chromatin accessibility in caMKL1-blocked cells Mice and cells. All mouse work was approved by the Institutional Animal Care (Fig. 6g). Importantly, this treatment alone resulted in the and Use Committee of Yale University. The reprogrammable mouse (R26rtTA; activation of endogenous pluripotency genes and led to the Col1a14F2A)60 (stock# 011004) was purchased from the Jackson Laboratory. The reprogrammable mice with reporter (R26rtTA;Col1a14F2A;Oct4GFP) were derived by emergence of mature pluripotent stem cells that were indepen- crossing reprogrammable mice with Oct4:GFP mice26,61. SRFf/fR26rtTATetO:Cre dent of Dox (Fig. 6h and Supplementary Fig. 9d). These data mouse line was derived by crossing the SRFf/f (stock# 006658) with tetO:Cre suggest that compromised chromatin accessibility driven by (stock# 006234). Sun2 knockout mouse has been described before6,9. MEFs with caMKL1 could largely explain their failure in activating mature different genetic background were all derived from E13.5 embryos. Granulocyte- pluripotency. Alleviating this inhibition, either by inhibiting Monocyte progenitors (GMPs) were collected from the bone marrow of mice with corresponding genotypes with the immune-phenotype of Lin-Sca-Kit + CD34 + MKL1-actin pathway directly or by inhibiting the CD16/32 + 18,61. Embryonic stem cells (ESCs) were derived from wild-type heterochromatin-forming enzymatic activity, restores chromatin C57BL/6 E3.5 embryos. accessibility and allows pluripotency activation. iPS cell induction. For reprogrammable MEFs, reprogramming was induced by adding Doxycycline (Dox) to the ESC culture medium with a final concentration of Discussion 2 μg/mL. For overexpression or knockdown experiments during reprogramming, In this report, we demonstrate that chromatin accessibility and viruses were generally transduced or co-transduced one day before changing the medium with or without Dox. The day when Dox is on is day 0. For all MEF mature pluripotency activation subject to negative modulation by reprogramming experiments, medium was changed every other day. On day 4, the ubiquitously expressed transcriptional co-activator MKL1 and reprogramming MEFs were trypsinized and seeded onto feeder layer with the its transcriptional product actin cytoskeleton. High MKL1 activity number of cells indicated. Unless otherwise indicated, AP staining was generally elevates the actin cytoskeleton, which connects with the nucleus performed on day 10 or day 12, depending on the size of the colonies. For reprogramming involving more than 12 days, reprogramming cells were passaged via the LINC complex. Sustained MKL1 activity results in and replated onto feeders at indicated time points. In selected cases (Sun1/2 reduced nuclear volume and globally reduced chromatin acces- knockdown, lamin A/C overexpression) where Oct4:GFP+ colonies could appear sibility. A smaller nuclear space could presumably reduce the without cell passaging, vitamin C was added into the culture medium starting on spatial distance separating distinct genomic loci, thereby expo- day 4. For GMPs, reprogramming cells were seeded onto feeder layer and mon- 49 itored for iPS colony formation without changing medium during the entire nentially increasing their contact probability or facilitating process. gene-silencing maintenance by PRC250. Such a model would favor the stabilization of a given chromatin conformation, rather Blocked cell maintenance and rescue. Blocked cells were cultured on feeder than its reorganization. This could potentially explain why the layers with ESC medium in the presence of 2 μg/mL Dox. For rescue experiments, activation, but less so maintenance, of mature pluripotency is cells were seeded onto gelatin-coated plates one day before viral transduction. Cells exquisitely sensitive to MKL1 activity. were incubated with viral medium supplemented with 5 μg/mL polybrene (Milli- Our work shifts the appreciation for the actin cytoskeleton- pore), and centrifuged at 2500 rpm, RT for 90 min. Cells were then incubated in the medium containing viral particles in 37 °C incubator for another 24 h. After that, LINC system, from an architectural network simply providing cells were trypsinized and replated onto feeder layer, cultured in ESC medium with nuclear structural support to an active contributor shaping the Dox to monitor rescuing activity. For rescue experiments with small molecules, epigenome. Our data collectively demonstrate that key compo- drugs were added to the medium, and cells were passaged every other day. ROCKi (Y-27632, CalBiochem, 688000) was used at 10 μM, CK-666 (Sigma, SML-0006) nents of the mechano-transduction machinery could restrict μ μ nuclear dynamics on a global level, revealing that the nuclear was used at 50 M, and UNC0642 (Sigma, SML-1037) was used at 1 M. process essential for cell fate reprogramming is modulated by mechanisms beyond the nuclear-localized epigenetic enzymes. Plasmids. MKL1-GFP construct was generated by PCR amplifying MKL1 from pcDNA5/TO-MKL162, replaced the IRES sequence and fused with the EGFP in The depiction of actin-LINC in restricting chromatin accessibility MIG-R1. Constitutively active MKL1 (caMKL1) was generated by deleting the first 51 could explain how certain kinases of cytoskeletal dynamics or 240 nucleotides (NM 001082536) encoding the RPEL domain and adding the SV40 actin-targeting shRNAs promoted reprograming52 and offer NLS before the remaining sequence. RhoA constructs were purchased from Addgene (12965, 12967, 12968)36 and subcloned into the pMSCV-IRES-mCherry mechanistic insights for how to integrate biophysical cues into 63 53 FP backbone (52114). LMNA/C construct was purchased from addgene (17662) cell fate engineering . Our work demonstrates that the cytos- and subcloned into the pMSCV-IRES-mCherry FP backbone, or fused with a keletal changes accompanying reprogramming are not merely mCherry at the N-terminus and subcloned into the pMSCV backbone. H1f0 was passive phenotypic changes, consequent to changes in chromatin subcloned into the pMSCV backbone15, with a mCherry fusion. Nesprin-2G remodeling. Instead, these changes could pose significant effects headless and tension sensor was obtained from Addgene45. The coding sequence of headless and tension sensor were subcloned into the pFUW lentiviral backbones toward chromatin remodeling. By sustaining the function of a key under the control of the TetO inducible promoter. MKL1 shRNA construct was molecular regulator for actin cytoskeleton, we uncovered and generated by inserting the short hairpin sequence into the lentiviral backbone psi- reenacted a molecular mechanism that is difficult to appreciate LVRU6MP (GeneCopia). Myh11, Fmn2, Arp2, Arp3, Sun1, and Sun2 shRNA otherwise, when cell states are in constant transitions during cell constructs were generated by inserting the short hairpin sequence into the lentiviral – fate reprogramming. backbone pLKO.1-blast. Each gene has 2 3 shRNA constructs, while only one is shown. Sequence information is provided in Supplementary Table 1. Since the MKL1-actin activity is ubiquitous, the proposed mechanism is likely relevant in other cellular contexts, including Chromatin immunoprecipitation (ChIP). Detailed method is provided in Sup- at the earlier stages of reprogramming or in ESC/iPSCs derived plementary Methods. For immunoprecipitation, Oct4 antibody (Cell Signaling, from other species. Intriguingly, one of the most distinct features 5677S) and SRF antibody (Active Motif, 61385, Santa Cruz, sc-335) were used at 6 between human and mouse pluripotent stem cells is their colony μg/ChIP sample, H3K4me3 antibody (Millipore, 07-473), H3K27ac antibody

10 NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6 ARTICLE

(Abcam, Ab4729), and H3K27me3 antibody (Millipore, 07-449) were used at 1 μg/ Nuclear volume measurement. Nuclear volume was measured with the ChIP sample. Imaris program. z-stack images were loaded to Imaris and converted to Imaris 3D image files (.ims). The surface detail was standardized at 0.002 mm for all images analyzed. The Surface Detail tool was used to control the surface of the nuclear RNA-seq and data analysis. RNA-seq libraries were prepared with TruSeq images, and he Background Subtraction tool was used to identify high Stranded mRNA Library Prep Kit (Illumina, RS-122-2101) following the manu- confidence nuclei. The Threshold (Background Subtraction) was kept at the facturer’s instructions. High-throughput sequencing was performed with the Illu- computer autogenerated average and the Split Touching Objects (Region Growing) mina HiSeq 2000 Sequencing System. For data analysis, the RNA-seq reads were option was enabled. The Seed Point Diameter was used to ensure individual mapped to mouse genome (mm10) with Bowtie2 in local mode, which allows the nuclei could be analyzed as distinct objects, since many nuclei partially overlap reads spanning the exon-exon junctions to get mapped to one of the two exons (especially with the Arp2 shRNA or SRF null cells). Only the nuclei that were (whichever gives the higher mapping score) independent of the transcriptome adequately separated from each other were recorded. The nuclei that overlapped annotation. Differentially expressed genes were identified by DESeq2 (v1.16.0) with each other and could not be segmented confidently were discarded. When the followed by cutting off with FDR-adjusted P value < 0.05 and fold changes > 2. MA surface was finalized, the tool Statistics was used to with the Detailed option and plot of differentially expressed genes was also done with the R software. Gene Set Specific Values tab selected. The Volume tab was used to generate the nuclear Enrichment Analysis (GSEA) was performed with the GSEA software. volumes. Reference gene sets were obtained from Polo et al.30. GO analysis of differentially expressed genes was performed with DAVID (https://david.ncifcrf.gov). FRAP assay. Blocked cells or control iPS cells were transduced with viruses expressing the H1f0-mCherry fusion protein. Images were acquired on a Leica SP5 ChIP-seq and data analysis. ChIPed DNA was tested for quality with BioAnalyzer ® confocal microscope. All experiments were done at 37 °C supplemented with 5% (Agilent) before library preparation. Libraries were constructed with ThruPLEX CO . For the FRAP experiments a pre-bleach image was acquired by averaging DNA-seq Kit (Rubicon Genomics) following the manufacturer’s instructions. 2 three consecutive images. Then a single spot of 2 μm×2μm region of interest High-throughput sequencing was performed with the Illumina HiSeq 2000 (ROI ) on the nuclei was bleached with the UV and 531 nm laser pulses at the Sequencing System. ChIP-seq reads were aligned to the mouse genome mm10 1 same time for 25 times lasting approximately 20 s, at 100% power of both lasers using Bowtie2 (v2.2.6) with default settings. For histone modifications, the resulting without scanning. For image collection, fluorescence was excited at 531 nm with BAM files were filtered to remove the reads that have the same sequence, and 20% power, and collected at 565–670 nm. 10 single section images were collected at converted to reads coverage files using the BamCoverage in DeepTools (v2.5.0). 0.78 s intervals, followed by 30 single section images collected at 2 s intervals. Reads coverage files were normalized as RPKM (Reads Per Kilobase of transcript An ROI signal on the fluorescent nuclei without bleaching (ROI ) and an ROI per Million mapped reads) and visualized in IGV (v2.3). Heatmap graphs were 2 signal of the background (ROI ) were collected for normalization. Plots were generated with DeepTools (v2.5.0). For transcription factors, filtered BAM files 3 generated by calculating I = [I − I ]/(I − I ), where I = (I − I )/(I were used for peak calling and differential peak analysis using MACS2 (v2.1.1). t 4 ave 4 t t.ROI1 t.ROI3 t.ROI2 − I ), I is the average I of the three pre-bleach images, and I is the first I Differential peaks were annotated with PAVIS (https://manticore.niehs.nih.gov/ t.ROI3 ave t 4 t post-bleach. pavis2/). GO analysis of differential peaks was performed with DAVID (https:// david.ncifcrf.gov). Motif analysis was performed by HOMER software. FRET analysis. High resolution live FRET imaging was performed on Nikon Eclipse Ti widefiled microscope equipped with a cooled charged-coupled device Nuclei preparation, DNase I digestion, and quantification. For nuclei pre- Cool SNAP HQ2 camera, using a ×100, 1.49 NA oil objective at 37 °C. Images were paration, live cells were suspended in RSB buffer (10 mM Tris-Cl, pH 7.4, 10 mM acquired using Micromanager software. Three sequential images with 500 ms NaCl, 3 mM MgCl2), and lysed by adding equal volume of RSB buffer supple- ® exposure time were acquired with the following filter combinations: donor (Teal) mented with 0.2% IGEPAL CA-630 (Sigma, I8896). Nuclei were spun down at channel with 460/20 (excitation filter-ex), T455lp (dichroic mirror-di) and 500/22 500 g, 4 °C, washed once with ice-cold DPBS, and resuspended in 1× DNase I (emission filter-em); FRET channel with 460/20 (ex), T455lp (di) and 535/30 (em); buffer in a concentration of 5 million nuclei/1 mL. Generally, aliquots of 100 k to and acceptor (Venus) channel with 492/18 (ex), T515lp (di) and 535/30 (em) filter 250 k nuclei were used for DNase I digestion (Roche, 04716728001). Nuclei were combinations. All filters and dichroic were purchased from Chroma Technology. digested with 0.1–0.4 U/μL of DNase I at 37 °C for 5 min, and stopped with equal For data analysis, donor leakage was determined from NIH3T3 cells transiently volume of stop buffer (50 mM Tris-HCl, pH 8.0, 100 mM NaCl, 100 mM EDTA, transfected with Vinculin-Teal, whereas acceptor cross excitation was obtained pH 8.0, 0.1% SDS). RNA was removed by adding RNase A (20 μg/mL) and from Vinculin-Venus transfected cells. For all the calculations, respective back- incubated at 37 °C for 30 min. Protein was subsequently digested by adding Pro- ground subtraction, illumination gradient, and pixel shift correction were per- teinase K (20 μg/mL) and incubated at 55 °C overnight. DNA was extracted by formed followed by three-point smoothening. The slope of pixel-wise donor or Phenol-Chloroform extraction. Reconstituted DNA was used for real-time PCR acceptor channel intensity versus FRET channel intensity gives leakage (x) or cross- analysis or agarose gel running. For quantification of the agarose gel image, the excitation (y) fraction, respectively. FRET map and pixel-wise FRET index for the intensity of DNA signals in each lane was quantified by GelAnalyzer2010a. Plots = − sensors were determined from were generated by calculating I (In Ib)/Itotal, where I is intensity, n is the location on the gel, b is background, and Itotal is sum of In-Ib. FRET index ¼ ½FRET channel À xðÞDonor channel ÀyðފAcceptor channel =½ŠAcceptor channel : ATAC-seq and analysis. Detailed method for ATAC-seq is provided in Supple- Mask for each cell was drawn manually based on the intensity around the metary Methods. For ATAC-seq analysis, sequence reads were trimmed using the nucleus. Mean FRET index per cell was calculated for each region within the trimmomatic software to remove nextera adapters and low quality bases at the mask65. beginning and end of the reads in paired end mode64. Reads were then aligned to the mouse mm10 genome sequence using the bwa mem mapper (arXiv:1303.3997 [q-bio.GN]). Reads were sorted and duplicates were marked using the Picard Reporting Summary. Further information on experimental design is available in software (http://broadinstitute.github.io/picard/). Sequence reads that aligned with the Nature Research Reporting Summary linked to this article. poor mapping quality to problematic black list regions or mitochondrial regions were removed. Peaks were called on reads with reads with insert size smaller than Data availability 250 bases using the MACS2 peak caller using the paired end mode -f BAMPE Datasets from RNA-seq, ChIP-seq and ATAC-seq are available under the accession option. Differentially bound peaks were identified using the DiffBind R package number GSE120399 in GEO [https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi? utilizing the Deseq2 method with parameters fragmentSize = 300 and bFullLi- acc=GSE120399]. All relevant data are available upon reasonable request. brarySize = TRUE (http://bioconductor.org/packages/release/bioc/html/DiffBind. html). Peaks were normalized by the total number of reads from all the samples. Received: 21 June 2018 Accepted: 1 March 2019 Antibody dilution for immunostaining. Lamin A/C antibody (Santa Cruz, sc376248) was used at 1:200 at 4 °C overnight.

Phalloidin staining, DAPI staining and nuclei analysis. For Phalloidin staining, Alexa Flour 660-conjugated Phalloidin (ThermoFisher, A22285) was used at RT for 30 min. DAPI (ThermoFisher, D1306) staining was performed at RT for 5 min. References Images were acquired by confocal microscopy (Leica SP5). Nuclear area and nuclear 1. Apostolou, E. & Hochedlinger, K. Chromatin dynamics during cellular intensity and nuclear circularity were measured with DAPI stained nuclei confocal reprogramming. Nature 502, 462–471 (2013). images in Image J, where circularity is defined as 4 × π (area perimeter−2). Fifty-five 2. Papp, B. & Plath, K. Epigenetics of reprogramming to induced pluripotency. to eighty nuclei were randomly selected for each genotype. For 3D reconstruction of Cell 152, 1324–1343 (2013). the nuclei, z-stack images were acquired on Leica SP5 confocal microscopy with z 3. Ungricht, R. & Kutay, U. Mechanisms and functions of nuclear envelope depth 0.25 μM. Images were stacked with the Leica software LAS X. remodelling. Nat. Rev. Mol. Cell Biol. 18, 229–245 (2017).

NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6

4. Sosa, B. A., Rothballer, A., Kutay, U. & Schwartz, T. U. LINC complexes form 34. Ridley, A. J. Rho GTPases and actin dynamics in membrane protrusions and by binding of three KASH peptides to domain interfaces of trimeric SUN vesicle trafficking. Trends. Cell Biol. 16, 522–529 (2006). proteins. Cell 149, 1035–1047 (2012). 35. Pellegrin, S. & Mellor, H. Actin stress fibres. J. Cell Sci. 120, 3491–3499 (2007). 5. Chang, W., Worman, H. J. & Gundersen, G. G. Accessorizing and anchoring 36. Subauste, M. C. et al. Rho family proteins modulate rapid apoptosis induced the LINC complex for multifunctionality. J. Cell Biol. 208,11–22 (2015). by cytotoxic T lymphocytes and Fas. J. Biol. Chem. 275, 9725–9733 (2000). 6. Stewart, R. M. et al. Nuclear-cytoskeletal linkages facilitate cross talk 37. Skau, C. T. et al. FMN2 makes perinuclear actin to protect nuclei during between the nucleus and intercellular adhesions. J. Cell Biol. 209, 403–418 confined migration and promote metastasis. Cell 167, 1571–1585 e1518 (2015). (2016). 7. Thiam, H. R. et al. Perinuclear Arp2/3-driven actin polymerization enables 38. Makhija, E., Jokhun, D. S. & Shivashankar, G. V. Nuclear deformability and nuclear deformation to facilitate cell migration through complex telomere dynamics are regulated by cell geometric constraints. Proc. Natl environments. Nat. Commun. 7, 10997 (2016). Acad. Sci. USA 113, E32–E40 (2016). 8. Zhang, X. et al. SUN1/2 and Syne/Nesprin-1/2 complexes connect centrosome 39. Versaevel, M., Grevesse, T. & Gabriele, S. Spatial coordination between cell to the nucleus during neurogenesis and neuronal migration in mice. Neuron and nuclear shape within micropatterned endothelial cells. Nat. Commun. 3, 64, 173–187 (2009). 671 (2012). 9. Lei, K. et al. SUN1 and SUN2 play critical but partially redundant roles in 40. Casella, J. F., Flanagan, M. D. & Lin, S. Cytochalasin D inhibits actin anchoring nuclei in skeletal muscle cells in mice. Proc. Natl Acad. Sci. USA polymerization and induces depolymerization of actin filaments formed 106, 10207–10212 (2009). during platelet shape change. Nature 293, 302–305 (1981). 10. Hatch, E. M. & Hetzer, M. W. Nuclear envelope rupture is induced by actin- 41. Miano, J. M. et al. Restricted inactivation of serum response factor to the based nucleus confinement. J. Cell Biol. 215,27–36 (2016). cardiovascular system. Proc. Natl Acad. Sci. USA 101, 17132–17137 (2004). 11. Lammerding, J. & Wolf, K. Nuclear envelope rupture: actin fibers are putting 42. Schratt, G. et al. Serum response factor is required for immediate-early gene the squeeze on the nucleus. J. Cell Biol. 215,5–8 (2016). activation yet is dispensable for proliferation of embryonic stem cells. Mol. 12. Shao, X., Li, Q., Mogilner, A., Bershadsky, A. D. & Shivashankar, G. V. Cell. Biol. 21, 2933–2943 (2001). Mechanical stimulation induces formin-dependent assembly of a perinuclear 43. Webster, M., Witkin, K. L. & Cohen-Fix, O. Sizing up the nucleus: nuclear actin rim. Proc. Natl Acad. Sci. USA 112, E2595–E2601 (2015). shape, size and nuclear-envelope assembly. J. Cell Sci. 122, 1477–1486 (2009). 13. Le, H. Q. et al. Mechanical regulation of transcription controls Polycomb- 44. Melcer, S. et al. Histone modifications and lamin A regulate chromatin protein mediated gene silencing during lineage commitment. Nat. Cell Biol. 18, dynamics in early embryonic stem cell differentiation. Nat. Commun. 3, 910 864–875 (2016). (2012). 14. Gaspar-Maia, A., Alajem, A., Meshorer, E. & Ramalho-Santos, M. Open 45. Arsenovic, P. T. et al. Nesprin-2G, a component of the nuclear LINC complex, chromatin in pluripotency and reprogramming. Nat. Rev. Mol. Cell Biol. 12, is subject to myosin-dependent tension. Biophys. J. 110,34–43 (2016). 36–47 (2011). 46. Krause, M., Te Riet, J. & Wolf, K. Probing the compressibility of tumor cell 15. Meshorer, E. et al. Hyperdynamic plasticity of chromatin proteins in nuclei by combined atomic force-confocal microscopy. Phys. Biol. 10, 065002 pluripotent embryonic stem cells. Dev. Cell. 10, 105–116 (2006). (2013). 16. Soufi, A., Donahue, G. & Zaret, K. S. Facilitators and impediments of the 47. Buenrostro, J. D., Giresi, P. G., Zaba, L. C., Chang, H. Y. & Greenleaf, W. J. pluripotency reprogramming factors’ initial engagement with the genome. Cell Transposition of native chromatin for fast and sensitive epigenomic profiling 151, 994–1004 (2012). of open chromatin, DNA-binding proteins and nucleosome position. Nat. 17. Sridharan, R. et al. Role of the murine reprogramming factors in the induction Methods 10, 1213–1218 (2013). of pluripotency. Cell 136, 364–377 (2009). 48. Liu, F. et al. Discovery of an in vivo chemical probe of the lysine 18. Guo, S. et al. Nonstochastic reprogramming from a privileged somatic cell methyltransferases G9a and GLP. J. Med. Chem. 56, 8931–8942 (2013). state. Cell 156, 649–662 (2014). 49. Wang, S. et al. Spatial organization of chromatin domains and compartments 19. Olson, E. N. & Nordheim, A. Linking actin dynamics and gene transcription in single chromosomes. Science 353, 598–602 (2016). to drive cellular motile functions. Nat. Rev. Mol. Cell Biol. 11, 353–365 50. Yuan, W. et al. Dense chromatin activates Polycomb repressive complex 2 to (2010). regulate H3 lysine 27 methylation. Science 337, 971–975 (2012). 20. Miano, J. M., Long, X. & Fujiwara, K. Serum response factor: master regulator 51. Sakurai, K. et al. Kinome-wide functional analysis highlights the role of of the actin cytoskeleton and contractile apparatus. American journal of cytoskeletal remodeling in somatic cell reprogramming. Cell. Stem. Cell. 14, physiology. Cell physiology 292, C70–C81 (2007). 523–534 (2014). 21. Miralles, F., Posern, G., Zaromytidou, A. I. & Treisman, R. Actin dynamics 52. Ikeda, T. et al. Srf destabilizes cellular identity by suppressing cell-type-specific control SRF activity by regulation of its coactivator MAL. Cell 113, 329–342 gene expression programs. Nat. Commun. 9, 1387 (2018). (2003). 53. Wong, S. Y., Soto, J. & Li, S. Biophysical regulation of cell reprogramming. 22. Vartiainen, M. K., Guettler, S., Larijani, B. & Treisman, R. Nuclear actin Curr Opin Chem Eng 15,95–101 (2017). regulates dynamic subcellular localization and activity of the SRF cofactor 54. Vitillo, L., Baxter, M., Iskender, B., Whiting, P. & Kimber, S. J. Integrin- MAL. Science 316, 1749–1752 (2007). associated focal adhesion kinase protects human embryonic stem cells from 23. Stadtfeld, M., Maherali, N., Borkent, M. & Hochedlinger, K. A apoptosis, detachment, and differentiation. Stem Cell Reports 7, 167–176 reprogrammable mouse strain from gene-targeted embryonic stem cells. Nat. (2016). Methods 7,53–55 (2010). 55. Collier, A. J. et al. Comprehensive cell surface protein profiling identifies 24. Ho, C. Y., Jaalouk, D. E., Vartiainen, M. K. & Lammerding, J. Lamin A/C and specific markers of human naive and primed pluripotent states. Cell. Stem. emerin regulate MKL1-SRF activity by modulating actin dynamics. Nature Cell. 20, 874–890 e877 (2017). 497, 507–511 (2013). 56. Guo, G. et al. Epigenetic resetting of human pluripotency. Development 144, 25. Lengner, C. J. et al. Oct4 expression is not required for mouse somatic stem 2748–2763 (2017). cell self-renewal. Cell. Stem. Cell. 1, 403–415 (2007). 57. Ezashi, T. et al. Derivation of induced pluripotent stem cells from pig somatic 26. Eminli, S. et al. Differentiation stage determines potential of hematopoietic cells. Proc. Natl Acad. Sci. USA 106, 10993–10998 (2009). cells for reprogramming into induced pluripotent stem cells. Nat. Genet. 41, 58. Helling, B. A. & Yang, I. V. Epigenetics in lung fibrosis: from pathobiology to 968–976 (2009). treatment perspective. Curr. Opin. Pulm. Med. 21, 454–462 (2015). 27. Megyola, C. M. et al. Dynamic migration and cell-cell interactions of early 59. Moran-Salvador, E. & Mann, J. Epigenetics and liver fibrosis. Cell. Mol. reprogramming revealed by high resolution time-lapse imaging. Stem Cells. Gastroenterol. Hepatol. 4, 125–134 (2017). https://doi.org/10.1002/stem.1323 (2013). 60. Carey, B. W., Markoulaki, S., Beard, C., Hanna, J. & Jaenisch, R. Single-gene 28. Esnault, C. et al. Rho-actin signaling to the MRTF coactivators dominates the transgenic mouse strains for reprogramming adult somatic cells. Nat. Methods immediate transcriptional response to serum in fibroblasts. Genes Dev. 28, 7,56–59 (2010). 943–958 (2014). 61. Megyola, C. M. et al. Dynamic migration and cell-cell interactions of early 29. Yue, F. et al. A comparative encyclopedia of DNA elements in the mouse reprogramming revealed by high-resolution time-lapse imaging. Stem Cells 31, genome. Nature 515, 355–364 (2014). 895–905 (2013). 30. Polo, J. M. et al. A molecular roadmap of reprogramming somatic cells into 62. Cheng, E. C. et al. Role for MKL1 in megakaryocytic maturation. Blood 113, iPS cells. Cell 151, 1617–1632 (2012). 2826–2834 (2009). 31. Chen, J. et al. H3K9 methylation is a barrier during somatic cell 63. Scaffidi, P. & Misteli, T. Lamin A-dependent misregulation of adult stem cells reprogramming into iPSCs. Nat. Genet. 45,34–42 (2013). associated with accelerated ageing. Nat. Cell Biol. 10, 452–459 (2008). 32. Mikkelsen, T. S. et al. Dissecting direct reprogramming through integrative 64. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trimmer for genomic analysis. Nature 454,49–55 (2008). Illumina sequence data. Bioinformatics 30, 2114–2120 (2014). 33. Tonge, P. D. et al. Divergent reprogramming routes lead to alternative stem- 65. Kumar, A. et al. Talin tension sensor reveals novel features of focal adhesion cell states. Nature 516, 192–197 (2014). force transmission and mechanosensitivity. J. Cell Biol. 213, 371–383 (2016).

12 NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-09636-6 ARTICLE

Acknowledgements Reprints and permission information is available online at http://npg.nature.com/ We thank Dr. Mistelli and Dr. Bustin and their lab members for sharing the reagents and reprintsandpermissions/ protocols for the FRAP assay. We thank Dr. Xinran Liu for assistance in the electron microscopy studies. This work is supported in part by the following funding: 15-RMB- Journal peer review information: Nature Communications thanks the anonymous YALE-03 and DP2 GM123507 (SG); NIH R01 DK094934 (DSK); NIH 1U54DK106857- reviewer(s) for their contribution to the peer review of this work. 02 (DSK/PGG), and NIH 2P01DK032094-29A1 (VPS/PGG). Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in fi Author contributions published maps and institutional af liations. X.H. designed, performed, analyzed most of the experiments and wrote the manuscript. S.G. supervised the project, designed experiments and wrote the manuscript. M.Z. per- formed the RNA-seq and ChIP-seq experiments. X.C. performed the ATAC-seq Open Access This article is licensed under a Creative Commons experiment. X.H. and Z.Z.L. analyzed the RNA-seq data and ChIP-seq data. V.P.S. Attribution 4.0 International License, which permits use, sharing, analyzed the ATAC-seq data. X.H. and A.K. performed the FRET experiments, and A.K. adaptation, distribution and reproduction in any medium or format, as long as you give analyzed the FRET data. J.W. analyzed the volume of all 3D nuclei images. L.M. provided appropriate credit to the original author(s) and the source, provide a link to the Creative the original caMKL1 construct. J.F.J. and E.C.R. provided animal resources. A.H., A.E., J. Commons license, and indicate if changes were made. The images or other third party C., C.C., P.G.G., J.L., M.S., M.C.K., D.S.K., and S.G. provided intellectual support. All the material in this article are included in the article’s Creative Commons license, unless authors reviewed the manuscript. indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory Additional information regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- licenses/by/4.0/. 019-09636-6.

Competing interests: The authors declare no competing interests. © The Author(s) 2019

NATURE COMMUNICATIONS | (2019) 10:1695 | https://doi.org/10.1038/s41467-019-09636-6 | www.nature.com/naturecommunications 13